SAR-DARTS: A CONVOLUTIONAL NEURAL ARCHITECTURE SEARCH FOR SAR SCENE CLASSIFICATION

Tian Zhou; Yinsheng Xu; Qihai Li; Baogui Qi; He Chen; Liang Chen

doi:10.1049/icp.2024.1189

SAR-DARTS: A CONVOLUTIONAL NEURAL ARCHITECTURE SEARCH FOR SAR SCENE CLASSIFICATION

Tian Zhou, Yinsheng Xu, Qihai Li, Baogui Qi^*, He Chen, Liang Chen

^*Corresponding author for this work

School of Information and Electronics

Research output: Contribution to journal › Conference article › peer-review

Abstract

In recent years, the collaborative service architecture of cloud computing and edge computing to reduce the demand for computing, storage and other capabilities of artificial intelligence methods has developed rapidly, but their application in synthetic aperture radar (SAR) image scene classification is still immature. In this paper, we propose a neural architecture search and pruning model for SAR scene classification. By introducing network morphisms and path regularization method, we can reduce the unfair competition between cells on the basis of expanding the search space. The model approach is applied to the FUSAR-Ship and OpenSARShip datasets for performance evaluation. The proposed method can achieve higher accuracy with fewer parameters.

Original language	English
Pages (from-to)	807-812
Number of pages	6
Journal	IET Conference Proceedings
Volume	2023
Issue number	47
DOIs	https://doi.org/10.1049/icp.2024.1189
Publication status	Published - 2023
Event	IET International Radar Conference 2023, IRC 2023 - Chongqing, China Duration: 3 Dec 2023 → 5 Dec 2023

Keywords

CONVOLUTIONAL NEURAL NETWORK
DIFFERENTIABLE ARCHITECTURE SEARCH
SCENE CLASSIFICATION
SYNTHETIC APERTURE RADAR

Access to Document

10.1049/icp.2024.1189

Cite this

@article{a6f30006d4d7492d9cf822e0b23a43db,

title = "SAR-DARTS: A CONVOLUTIONAL NEURAL ARCHITECTURE SEARCH FOR SAR SCENE CLASSIFICATION",

abstract = "In recent years, the collaborative service architecture of cloud computing and edge computing to reduce the demand for computing, storage and other capabilities of artificial intelligence methods has developed rapidly, but their application in synthetic aperture radar (SAR) image scene classification is still immature. In this paper, we propose a neural architecture search and pruning model for SAR scene classification. By introducing network morphisms and path regularization method, we can reduce the unfair competition between cells on the basis of expanding the search space. The model approach is applied to the FUSAR-Ship and OpenSARShip datasets for performance evaluation. The proposed method can achieve higher accuracy with fewer parameters.",

keywords = "CONVOLUTIONAL NEURAL NETWORK, DIFFERENTIABLE ARCHITECTURE SEARCH, SCENE CLASSIFICATION, SYNTHETIC APERTURE RADAR",

author = "Tian Zhou and Yinsheng Xu and Qihai Li and Baogui Qi and He Chen and Liang Chen",

note = "Publisher Copyright: {\textcopyright} The Institution of Engineering \& Technology 2023.; IET International Radar Conference 2023, IRC 2023 ; Conference date: 03-12-2023 Through 05-12-2023",

year = "2023",

doi = "10.1049/icp.2024.1189",

language = "English",

volume = "2023",

pages = "807--812",

journal = "IET Conference Proceedings",

issn = "2732-4494",

publisher = "Institution of Engineering and Technology",

number = "47",

}

TY - JOUR

T1 - SAR-DARTS

T2 - IET International Radar Conference 2023, IRC 2023

AU - Zhou, Tian

AU - Xu, Yinsheng

AU - Li, Qihai

AU - Qi, Baogui

AU - Chen, He

AU - Chen, Liang

N1 - Publisher Copyright: © The Institution of Engineering & Technology 2023.

PY - 2023

Y1 - 2023

N2 - In recent years, the collaborative service architecture of cloud computing and edge computing to reduce the demand for computing, storage and other capabilities of artificial intelligence methods has developed rapidly, but their application in synthetic aperture radar (SAR) image scene classification is still immature. In this paper, we propose a neural architecture search and pruning model for SAR scene classification. By introducing network morphisms and path regularization method, we can reduce the unfair competition between cells on the basis of expanding the search space. The model approach is applied to the FUSAR-Ship and OpenSARShip datasets for performance evaluation. The proposed method can achieve higher accuracy with fewer parameters.

AB - In recent years, the collaborative service architecture of cloud computing and edge computing to reduce the demand for computing, storage and other capabilities of artificial intelligence methods has developed rapidly, but their application in synthetic aperture radar (SAR) image scene classification is still immature. In this paper, we propose a neural architecture search and pruning model for SAR scene classification. By introducing network morphisms and path regularization method, we can reduce the unfair competition between cells on the basis of expanding the search space. The model approach is applied to the FUSAR-Ship and OpenSARShip datasets for performance evaluation. The proposed method can achieve higher accuracy with fewer parameters.

KW - CONVOLUTIONAL NEURAL NETWORK

KW - DIFFERENTIABLE ARCHITECTURE SEARCH

KW - SCENE CLASSIFICATION

KW - SYNTHETIC APERTURE RADAR

UR - http://www.scopus.com/inward/record.url?scp=85203173978&partnerID=8YFLogxK

U2 - 10.1049/icp.2024.1189

DO - 10.1049/icp.2024.1189

M3 - Conference article

AN - SCOPUS:85203173978

SN - 2732-4494

VL - 2023

SP - 807

EP - 812

JO - IET Conference Proceedings

JF - IET Conference Proceedings

IS - 47

Y2 - 3 December 2023 through 5 December 2023

ER -

SAR-DARTS: A CONVOLUTIONAL NEURAL ARCHITECTURE SEARCH FOR SAR SCENE CLASSIFICATION

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this